Remove carriage returns
This commit is contained in:
Nathan Sobo
parent
28b71cbc03
commit
dd6f8d20a3
3 changed files with 9721 additions and 9721 deletions
File diff suppressed because it is too large
Load diff
File diff suppressed because it is too large
Load diff
|
|
@ -1,227 +1,227 @@
|
||||||
use aho_corasick::{AhoCorasick, AhoCorasickBuilder};
|
use aho_corasick::{AhoCorasick, AhoCorasickBuilder};
|
||||||
use anyhow::Result;
|
use anyhow::Result;
|
||||||
use client::proto;
|
use client::proto;
|
||||||
use language::{char_kind, Rope};
|
use language::{char_kind, Rope};
|
||||||
use regex::{Regex, RegexBuilder};
|
use regex::{Regex, RegexBuilder};
|
||||||
use smol::future::yield_now;
|
use smol::future::yield_now;
|
||||||
use std::{
|
use std::{
|
||||||
io::{BufRead, BufReader, Read},
|
io::{BufRead, BufReader, Read},
|
||||||
ops::Range,
|
ops::Range,
|
||||||
sync::Arc,
|
sync::Arc,
|
||||||
};
|
};
|
||||||
|
|
||||||
#[derive(Clone)]
|
#[derive(Clone)]
|
||||||
pub enum SearchQuery {
|
pub enum SearchQuery {
|
||||||
Text {
|
Text {
|
||||||
search: Arc<AhoCorasick<usize>>,
|
search: Arc<AhoCorasick<usize>>,
|
||||||
query: Arc<str>,
|
query: Arc<str>,
|
||||||
whole_word: bool,
|
whole_word: bool,
|
||||||
case_sensitive: bool,
|
case_sensitive: bool,
|
||||||
},
|
},
|
||||||
Regex {
|
Regex {
|
||||||
regex: Regex,
|
regex: Regex,
|
||||||
query: Arc<str>,
|
query: Arc<str>,
|
||||||
multiline: bool,
|
multiline: bool,
|
||||||
whole_word: bool,
|
whole_word: bool,
|
||||||
case_sensitive: bool,
|
case_sensitive: bool,
|
||||||
},
|
},
|
||||||
}
|
}
|
||||||
|
|
||||||
impl SearchQuery {
|
impl SearchQuery {
|
||||||
pub fn text(query: impl ToString, whole_word: bool, case_sensitive: bool) -> Self {
|
pub fn text(query: impl ToString, whole_word: bool, case_sensitive: bool) -> Self {
|
||||||
let query = query.to_string();
|
let query = query.to_string();
|
||||||
let search = AhoCorasickBuilder::new()
|
let search = AhoCorasickBuilder::new()
|
||||||
.auto_configure(&[&query])
|
.auto_configure(&[&query])
|
||||||
.ascii_case_insensitive(!case_sensitive)
|
.ascii_case_insensitive(!case_sensitive)
|
||||||
.build(&[&query]);
|
.build(&[&query]);
|
||||||
Self::Text {
|
Self::Text {
|
||||||
search: Arc::new(search),
|
search: Arc::new(search),
|
||||||
query: Arc::from(query),
|
query: Arc::from(query),
|
||||||
whole_word,
|
whole_word,
|
||||||
case_sensitive,
|
case_sensitive,
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn regex(query: impl ToString, whole_word: bool, case_sensitive: bool) -> Result<Self> {
|
pub fn regex(query: impl ToString, whole_word: bool, case_sensitive: bool) -> Result<Self> {
|
||||||
let mut query = query.to_string();
|
let mut query = query.to_string();
|
||||||
let initial_query = Arc::from(query.as_str());
|
let initial_query = Arc::from(query.as_str());
|
||||||
if whole_word {
|
if whole_word {
|
||||||
let mut word_query = String::new();
|
let mut word_query = String::new();
|
||||||
word_query.push_str("\\b");
|
word_query.push_str("\\b");
|
||||||
word_query.push_str(&query);
|
word_query.push_str(&query);
|
||||||
word_query.push_str("\\b");
|
word_query.push_str("\\b");
|
||||||
query = word_query
|
query = word_query
|
||||||
}
|
}
|
||||||
|
|
||||||
let multiline = query.contains("\n") || query.contains("\\n");
|
let multiline = query.contains("\n") || query.contains("\\n");
|
||||||
let regex = RegexBuilder::new(&query)
|
let regex = RegexBuilder::new(&query)
|
||||||
.case_insensitive(!case_sensitive)
|
.case_insensitive(!case_sensitive)
|
||||||
.multi_line(multiline)
|
.multi_line(multiline)
|
||||||
.build()?;
|
.build()?;
|
||||||
Ok(Self::Regex {
|
Ok(Self::Regex {
|
||||||
regex,
|
regex,
|
||||||
query: initial_query,
|
query: initial_query,
|
||||||
multiline,
|
multiline,
|
||||||
whole_word,
|
whole_word,
|
||||||
case_sensitive,
|
case_sensitive,
|
||||||
})
|
})
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn from_proto(message: proto::SearchProject) -> Result<Self> {
|
pub fn from_proto(message: proto::SearchProject) -> Result<Self> {
|
||||||
if message.regex {
|
if message.regex {
|
||||||
Self::regex(message.query, message.whole_word, message.case_sensitive)
|
Self::regex(message.query, message.whole_word, message.case_sensitive)
|
||||||
} else {
|
} else {
|
||||||
Ok(Self::text(
|
Ok(Self::text(
|
||||||
message.query,
|
message.query,
|
||||||
message.whole_word,
|
message.whole_word,
|
||||||
message.case_sensitive,
|
message.case_sensitive,
|
||||||
))
|
))
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn to_proto(&self, project_id: u64) -> proto::SearchProject {
|
pub fn to_proto(&self, project_id: u64) -> proto::SearchProject {
|
||||||
proto::SearchProject {
|
proto::SearchProject {
|
||||||
project_id,
|
project_id,
|
||||||
query: self.as_str().to_string(),
|
query: self.as_str().to_string(),
|
||||||
regex: self.is_regex(),
|
regex: self.is_regex(),
|
||||||
whole_word: self.whole_word(),
|
whole_word: self.whole_word(),
|
||||||
case_sensitive: self.case_sensitive(),
|
case_sensitive: self.case_sensitive(),
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn detect<T: Read>(&self, stream: T) -> Result<bool> {
|
pub fn detect<T: Read>(&self, stream: T) -> Result<bool> {
|
||||||
if self.as_str().is_empty() {
|
if self.as_str().is_empty() {
|
||||||
return Ok(false);
|
return Ok(false);
|
||||||
}
|
}
|
||||||
|
|
||||||
match self {
|
match self {
|
||||||
Self::Text { search, .. } => {
|
Self::Text { search, .. } => {
|
||||||
let mat = search.stream_find_iter(stream).next();
|
let mat = search.stream_find_iter(stream).next();
|
||||||
match mat {
|
match mat {
|
||||||
Some(Ok(_)) => Ok(true),
|
Some(Ok(_)) => Ok(true),
|
||||||
Some(Err(err)) => Err(err.into()),
|
Some(Err(err)) => Err(err.into()),
|
||||||
None => Ok(false),
|
None => Ok(false),
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
Self::Regex {
|
Self::Regex {
|
||||||
regex, multiline, ..
|
regex, multiline, ..
|
||||||
} => {
|
} => {
|
||||||
let mut reader = BufReader::new(stream);
|
let mut reader = BufReader::new(stream);
|
||||||
if *multiline {
|
if *multiline {
|
||||||
let mut text = String::new();
|
let mut text = String::new();
|
||||||
if let Err(err) = reader.read_to_string(&mut text) {
|
if let Err(err) = reader.read_to_string(&mut text) {
|
||||||
Err(err.into())
|
Err(err.into())
|
||||||
} else {
|
} else {
|
||||||
Ok(regex.find(&text).is_some())
|
Ok(regex.find(&text).is_some())
|
||||||
}
|
}
|
||||||
} else {
|
} else {
|
||||||
for line in reader.lines() {
|
for line in reader.lines() {
|
||||||
let line = line?;
|
let line = line?;
|
||||||
if regex.find(&line).is_some() {
|
if regex.find(&line).is_some() {
|
||||||
return Ok(true);
|
return Ok(true);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
Ok(false)
|
Ok(false)
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
pub async fn search(&self, rope: &Rope) -> Vec<Range<usize>> {
|
pub async fn search(&self, rope: &Rope) -> Vec<Range<usize>> {
|
||||||
const YIELD_INTERVAL: usize = 20000;
|
const YIELD_INTERVAL: usize = 20000;
|
||||||
|
|
||||||
if self.as_str().is_empty() {
|
if self.as_str().is_empty() {
|
||||||
return Default::default();
|
return Default::default();
|
||||||
}
|
}
|
||||||
|
|
||||||
let mut matches = Vec::new();
|
let mut matches = Vec::new();
|
||||||
match self {
|
match self {
|
||||||
Self::Text {
|
Self::Text {
|
||||||
search, whole_word, ..
|
search, whole_word, ..
|
||||||
} => {
|
} => {
|
||||||
for (ix, mat) in search
|
for (ix, mat) in search
|
||||||
.stream_find_iter(rope.bytes_in_range(0..rope.len()))
|
.stream_find_iter(rope.bytes_in_range(0..rope.len()))
|
||||||
.enumerate()
|
.enumerate()
|
||||||
{
|
{
|
||||||
if (ix + 1) % YIELD_INTERVAL == 0 {
|
if (ix + 1) % YIELD_INTERVAL == 0 {
|
||||||
yield_now().await;
|
yield_now().await;
|
||||||
}
|
}
|
||||||
|
|
||||||
let mat = mat.unwrap();
|
let mat = mat.unwrap();
|
||||||
if *whole_word {
|
if *whole_word {
|
||||||
let prev_kind = rope.reversed_chars_at(mat.start()).next().map(char_kind);
|
let prev_kind = rope.reversed_chars_at(mat.start()).next().map(char_kind);
|
||||||
let start_kind = char_kind(rope.chars_at(mat.start()).next().unwrap());
|
let start_kind = char_kind(rope.chars_at(mat.start()).next().unwrap());
|
||||||
let end_kind = char_kind(rope.reversed_chars_at(mat.end()).next().unwrap());
|
let end_kind = char_kind(rope.reversed_chars_at(mat.end()).next().unwrap());
|
||||||
let next_kind = rope.chars_at(mat.end()).next().map(char_kind);
|
let next_kind = rope.chars_at(mat.end()).next().map(char_kind);
|
||||||
if Some(start_kind) == prev_kind || Some(end_kind) == next_kind {
|
if Some(start_kind) == prev_kind || Some(end_kind) == next_kind {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
matches.push(mat.start()..mat.end())
|
matches.push(mat.start()..mat.end())
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
Self::Regex {
|
Self::Regex {
|
||||||
regex, multiline, ..
|
regex, multiline, ..
|
||||||
} => {
|
} => {
|
||||||
if *multiline {
|
if *multiline {
|
||||||
let text = rope.to_string();
|
let text = rope.to_string();
|
||||||
for (ix, mat) in regex.find_iter(&text).enumerate() {
|
for (ix, mat) in regex.find_iter(&text).enumerate() {
|
||||||
if (ix + 1) % YIELD_INTERVAL == 0 {
|
if (ix + 1) % YIELD_INTERVAL == 0 {
|
||||||
yield_now().await;
|
yield_now().await;
|
||||||
}
|
}
|
||||||
|
|
||||||
matches.push(mat.start()..mat.end());
|
matches.push(mat.start()..mat.end());
|
||||||
}
|
}
|
||||||
} else {
|
} else {
|
||||||
let mut line = String::new();
|
let mut line = String::new();
|
||||||
let mut line_offset = 0;
|
let mut line_offset = 0;
|
||||||
for (chunk_ix, chunk) in rope.chunks().chain(["\n"]).enumerate() {
|
for (chunk_ix, chunk) in rope.chunks().chain(["\n"]).enumerate() {
|
||||||
if (chunk_ix + 1) % YIELD_INTERVAL == 0 {
|
if (chunk_ix + 1) % YIELD_INTERVAL == 0 {
|
||||||
yield_now().await;
|
yield_now().await;
|
||||||
}
|
}
|
||||||
|
|
||||||
for (newline_ix, text) in chunk.split('\n').enumerate() {
|
for (newline_ix, text) in chunk.split('\n').enumerate() {
|
||||||
if newline_ix > 0 {
|
if newline_ix > 0 {
|
||||||
for mat in regex.find_iter(&line) {
|
for mat in regex.find_iter(&line) {
|
||||||
let start = line_offset + mat.start();
|
let start = line_offset + mat.start();
|
||||||
let end = line_offset + mat.end();
|
let end = line_offset + mat.end();
|
||||||
matches.push(start..end);
|
matches.push(start..end);
|
||||||
}
|
}
|
||||||
|
|
||||||
line_offset += line.len() + 1;
|
line_offset += line.len() + 1;
|
||||||
line.clear();
|
line.clear();
|
||||||
}
|
}
|
||||||
line.push_str(text);
|
line.push_str(text);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
matches
|
matches
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn as_str(&self) -> &str {
|
pub fn as_str(&self) -> &str {
|
||||||
match self {
|
match self {
|
||||||
Self::Text { query, .. } => query.as_ref(),
|
Self::Text { query, .. } => query.as_ref(),
|
||||||
Self::Regex { query, .. } => query.as_ref(),
|
Self::Regex { query, .. } => query.as_ref(),
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn whole_word(&self) -> bool {
|
pub fn whole_word(&self) -> bool {
|
||||||
match self {
|
match self {
|
||||||
Self::Text { whole_word, .. } => *whole_word,
|
Self::Text { whole_word, .. } => *whole_word,
|
||||||
Self::Regex { whole_word, .. } => *whole_word,
|
Self::Regex { whole_word, .. } => *whole_word,
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn case_sensitive(&self) -> bool {
|
pub fn case_sensitive(&self) -> bool {
|
||||||
match self {
|
match self {
|
||||||
Self::Text { case_sensitive, .. } => *case_sensitive,
|
Self::Text { case_sensitive, .. } => *case_sensitive,
|
||||||
Self::Regex { case_sensitive, .. } => *case_sensitive,
|
Self::Regex { case_sensitive, .. } => *case_sensitive,
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
pub fn is_regex(&self) -> bool {
|
pub fn is_regex(&self) -> bool {
|
||||||
matches!(self, Self::Regex { .. })
|
matches!(self, Self::Regex { .. })
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue